Effect of Delaying Treatment on Efficacy of Ciprofloxacin and Levofloxacin in the African Green Monkey Model of Pneumonic Plague

Abstract

Background

Ciprofloxacin and levofloxacin, 2 fluoroquinolone antimicrobials, are ≥90% effective for the treatment of inhalational plague when administered within 2–6 hours of fever onset in African green monkeys (AGM). Based on data in the AGM model, these antimicrobials were approved under the Food and Drug Administration’s Animal Efficacy Rule. However, that data did not address the issue of how long treatment with these antimicrobials would remain effective after fever onset.

Methods

The AGM model of pneumonic plague was used to explore the effect of delaying treatment with ciprofloxacin and levofloxacin on efficacy. In 2 studies, AGMs were challenged with inhaled lethal doses of Yersinia pestis. Treatment with ciprofloxacin and levofloxacin was initiated from 0 to up to 30 hours after fever onset.

Results

Challenged animals all developed fever within 78 hours and were treated with ciprofloxacin (n = 27) or levofloxacin (n = 29) at various predetermined time points postfever. When administered 10 hours after fever onset, 10 days of ciprofloxacin and levofloxacin treatment remained very effective (90 or 100%, respectively). The efficacy of both antimicrobials declined as treatment initiation was further delayed. Statistical analyses estimated the treatment delay times at which half of the AGMs were no longer expected to survive as 19.7 hours for ciprofloxacin and 26.5 hours for levofloxacin.

Conclusions

This study demonstrates that there is a narrow window following fever onset during which ciprofloxacin and levofloxacin are fully effective treatment options for pneumonic plague in AGMs. Since the timing of disease is similar in humans and AGMs, these AGM data are reasonably likely to predict response times for treatment in humans.

Delaying the treatment of pneumonic plague with ciprofloxacin and levofloxacin by 20–27 hours after fever onset substantially reduces their efficacy in African green monkeys.

The Gram-negative bacillus Yersinia pestis is the causative agent of plague. Y. pestis infection is often lethal in the absence of treatment and was responsible for millions of deaths during the third pandemic, which spread from China to ports worldwide via ships [1]. Because of its lethality and potential use as a biological weapon, Y. pestis is a Tier 1 Select Agent.

Recently, the Food and Drug Administration (FDA) approved 2 fluoroquinolone antimicrobials, ciprofloxacin and levofloxacin, for the treatment of pneumonic plague, based on efficacy data obtained in an African green monkey (AGM) disease model. Data presented to the FDA demonstrated that ciprofloxacin and levofloxacin were efficacious when treatment was initiated within 2–6 hours of fever onset [2, 3]. In real-world situations, however, treatment of humans with pneumonic plague is not likely to begin at the time of fever onset. The Anti-Infective Drug Advisory Committee that recommended the approval of ciprofloxacin and levofloxacin discussed the timing of treatment initiation. Therefore, information on antimicrobial efficacy by timing of treatment would help improve our understanding of the disease progression, the determination of the optimal window for initiation of treatment, and outbreak response planning.

In this work, the relationship between the time of treatment initiation and the efficacy of these antimicrobials was explored in the AGM model in 2 independent studies. An important goal of this work was to provide an estimate of how long after the onset of fever these antimicrobials could be expected to be effective in humans.

METHODS

Aerosol Challenges

Y. pestis strain CO92 was used to challenge AGMs. CO92 was isolated from a lethal human infection [4]. Challenge material was harvested following the growth of CO92 on tryptose blood agar base slants, aerosolized in a Collison nebulizer, and administered to AGMs via inhalation in a head-only aerosol exposure chamber following brief anesthesia, as described elsewhere in this issue [2]. The target challenge dose was 100 LD₅₀ (lethal dose 50 and is the challenge dose that results in 50% of the challenged animals dying in the absence of treatment) and actual doses were quantified as described (1 LD₅₀ equals a challenge dose of ~343 colony-forming units) [2].

Fever Determination

Body temperature was determined by the use of surgically implanted telemetry units, as described elsewhere in this issue [2, 3]. For each AGM, the circadian-adjusted body temperature was determined by telemetry readings prior to the challenge. Fever was defined as a body temperature 1.5℃ above the normal circadian-adjusted body temperature.

Ciprofloxacin Study Design

This study was conducted in 2 phases, with 19–20 AGMs in each. AGMs in the first phase were randomized into treatment and control groups that were treated at 0, 5, 10, or 15 hours postfever. AGMs in the second phase were randomized after the interim data analysis from the first phase, to postfever treatment times of 12, 20, or 28 hours. Actual times of fever and treatment were recorded and used in the statistical analysis. Ciprofloxacin doses were administered every 12 hours for 10 days, except in cases of premature death, which resulted in fewer doses being administered. Each dose of 15 mg/kg ciprofloxacin or 0.5% dextrose (placebo) was administered over 60 minutes via a venous access port. Ciprofloxacin was manufactured by Bayer Health Care Pharmaceuticals. Survival was the primary endpoint and animals were observed for 21 days postchallenge, at which time all animals were euthanized.

Levofloxacin Study

A total of 39 animals were randomly assigned to either the treatment or control group, as well as to treatment initiation at 0, 8, 16, or 24 hours postfever; the actual times of fever and treatment were recorded and used in statistical analyses. Levofloxacin at 10 mg/kg/day was split into 2 doses, with the first dose of 8 mg/kg followed 12 hours later by a dose of 2 mg/kg, in order to approximate the pharmacokinetic profile of a once-daily human dose. Levofloxacin or placebo (7.5 mL/kg of a 5% dextrose solution) was administered for 30 minutes via a surgically implanted indwelling catheter for 10 days, except in the cases of premature death, which resulted in fewer doses being administered. Levofloxacin was manufactured by US Pharmacopeia. Survival was the primary endpoint and the animals were observed until scheduled euthanasia on Day 28 postchallenge.

RESULTS

Ciprofloxacin Study

In this study, 39 AGMs were challenged with the CO92 strain of Y. pestis via an inhalational route (Table 1). Challenge doses ranged from 111 LD₅₀s to 521 LD₅₀s. The average challenge dose was 245 LD₅₀s. AGMs were randomized to be treated with ciprofloxacin (or placebo) at various predetermined times after fever onset. The time of fever onset ranged from 47.8 to 76.8 hours (data not shown). The time to fever onset was not substantially correlated with the challenge dose (Pearson’s correlation r = −0.13; 95% confidence interval [CI], −.43 to .19).

Table 1.

Timing of Ciprofloxacin or Levofloxacin Treatment and Outcome of African Green Monkeys Infected with Yersinia pestis by Inhalation

Animal Number	Sex	Antimicrobial	Time Between Fever Onset and Initiation of Antimicrobial Treatmenta	Died Before Scheduled Euthanasiab
1	M	Ciprofloxacin	.5 hours	No
2	F	Ciprofloxacin	.6 hours	No
3	F	Ciprofloxacin	.6 hours	No
4	M	Ciprofloxacin	.7 hours	No
5	F	Ciprofloxacin	5.0 hours	No
6	M	Ciprofloxacin	5.1 hours	No
7	M	Ciprofloxacin	5.2 hours	No
8	F	Ciprofloxacin	5.2 hours	No
9	F	Ciprofloxacin	7.5 hours	Yes
10	F	Ciprofloxacin	10.0 hours	No
11	M	Ciprofloxacin	10.2 hours	Yes
12	M	Ciprofloxacin	10.2 hours	No
13	F	Ciprofloxacin	10.3 hours	No
14	M	Ciprofloxacin	12.0 hours	No
15	F	Ciprofloxacin	12.0 hours	No
16	F	Ciprofloxacin	12.2 hours	Yes
17	M	Ciprofloxacin	12.4 hours	No
18	M	Ciprofloxacin	15.0 hours	No
19	F	Ciprofloxacin	15.1 hours	No
20	F	Ciprofloxacin	15.2 hours	No
21	F	Ciprofloxacin	20.1 hours	Yes
22	F	Ciprofloxacin	20.1 hours	Yes
23	M	Ciprofloxacin	20.4 hours	No
24	M	Ciprofloxacin	20.6 hours	Yes
25	F	Ciprofloxacin	28.1 hours	Yes
26	M	Ciprofloxacin	28.5 hours	No
27	F	Ciprofloxacin	28.5 hours	Yes
28	M	N/A	Died before any treatmentc	Yes
29	M	N/A	Died before any treatmentd	Yes
30	F	N/A	Died before any treatmente	Yes
31	M	N/A	Died before any treatmentf	Yes
32 controlg	M	Placebo	.5 hours	Yes
33 controlg	M	Placebo	.6 hours	Yes
34 controlg	F	Placebo	.6 hours	Yes
35 controlg	F	Placebo	.6 hours	Yes
36 controlg	M	Placebo	.7 hours	Yes
37 controlg	F	Placebo	.8 hours	Yes
38 controlg	M	Placebo	1.1 hours	Yes
39 controlg	F	Placebo	1.4 hours	Yes
40	F	Levofloxacin	.7 hours	No
41	M	Levofloxacin	2.5 hours	No
42	M	Levofloxacin	2.5 hours	No
43	M	Levofloxacin	2.9 hours	No
44	F	Levofloxacin	3.9 hours	No
45	F	Levofloxacin	7.6 hours	No
46	F	Levofloxacin	10.3 hours	No
47	M	Levofloxacin	10.5 hours	No
48	F	Levofloxacin	12.8 hours	No
49	M	Levofloxacin	16.0 hours	No
50	F	Levofloxacin	16.2 hours	No
51	M	Levofloxacin	19.4 hours	No
52	M	Levofloxacin	19.8 hours	No
53	F	Levofloxacin	19.9 hours	No
54	F	Levofloxacin	20.0 hours	No
55	M	Levofloxacin	20.1 hours	No
56	M	Levofloxacin	21.2 hours	No
57	M	Levofloxacin	22.0 hours	No
58	F	Levofloxacin	24.0 hours	Yes
59	F	Levofloxacin	24.1 hours	No
60	F	Levofloxacin	24.3 hours	Yes
61	F	Levofloxacin	25.9 hours	No
62	M	Levofloxacin	26.0 hours	No
63	F	Levofloxacin	26.9 hours	No
64	M	Levofloxacin	27.4 hours	Yes
65	F	Levofloxacin	27.5 hours	No
66	M	Levofloxacin	28.5 hours	Yes
67	M	Levofloxacin	29.5 hours	Yes
68	M	Levofloxacin	29.8 hours	Yes
69	F	Levofloxacin	−1.9 hours	No
70	M	Levofloxacin	−5.4 hours	Removed from studyh
71	F	N/A	Died before any treatmenti	Yes
72 controlj	F	Placebo	1.6 hours	Yes
73 controlj	M	Placebo	10.3 hours	Yes
74 controlj	F	Placebo	11.1 hours	Yes
75 controlj	F	Placebo	16.7 hours	Yes
76 controlj	M	Placebo	21.4 hours	Yes
77 controlj	M	Placebo	24.0 hours	Yes
78 controlj	M	Placebo	−4.8 hours	Yes
79 controlj	F	Placebo	Died before any treatment	Yes

Abbreviation: AGM, African green monkey.

^aThe time from Yersinia pestis challenge to fever onset ranged from 47.8 to 76.8 hours, with an average of 57.7 hours for AGMs treated with ciprofloxacin. For AGMs treated with levofloxacin, the time from challenge to fever onset ranged from 52.5 to 78.4 hours, with an average of 55.6 hours.

^bFor ciprofloxacin-treated animals, the scheduled euthanasia was 500 hours postchallenge. For levofloxacin-treated animals, the scheduled euthanasia was 672 hours postchallenge. Animals that did not survive until the time of scheduled euthanasia were either found deceased or found to be moribund and euthanized ahead of schedule.

^cAnimal 28 was scheduled to be treated 15 hours after fever onset and died 10.1 hours after fever onset.

^dAnimal 29 was scheduled to be treated 20 hours after fever onset and died 19 hours after fever onset.

^eAnimal 30 was scheduled to be treated 28 hours after fever onset and died 21 hours after fever onset.

^fAnimal 31 was scheduled to be treated 28 hours after fever onset and died 22.9 hours after fever onset.

^gControl animals received 60 minutes of 7.5 mL/kg of a solution of 5% dextrose and died an average of 20 hours after fever onset, with a range of 7–36 hours.

^hAnimal 70 was removed from the study because of a premature fever call.

ⁱAnimal 71 was scheduled to be treated 24 hours after fever and died 20 hours after fever.

^jControl animals received an equivalent volume of a solution of 5% dextrose and died an average of 45 hours after fever with a range of 18–76 hours.

All 8 placebo animals died within 40 hours of the time of fever. The correlation of time to death with the challenge dose was r = −0.37, with a large CI (95% CI, −.85 to .45). In addition, 4 AGMs died prior to their first scheduled ciprofloxacin dose (Table 1). The remaining 27 AGMs were treated with ciprofloxacin at various intervals after the onset of fever (Table 1). The 27 animals treated with ciprofloxacin demonstrated that a delay in treatment results in a reduction in efficacy (Table 2). The statistical analysis revealed that delaying ciprofloxacin treatment by 19.7 hours results in a treatment efficacy of 50%, as compared to the higher level of efficacy observed when treatment is initiated at or near the time of fever onset (Table 2). Figure 1 depicts the clinical trajectories of the 27 individual ciprofloxacin-treated AGMs and a curve fit, providing an estimate of the rate of efficacy decay.

Table 2.

Overall Survival of African Green Monkeys Infected with Yersinia pestis by Inhalation and Treated with Ciprofloxacin or Levofloxacin, by Treatment Window

Antimicrobial	Number of Animals that Survived/ Total Number Treated 0–10 Hours After Fever Onset (%)	Number of Animals that Survived/ Total Number Treated 10.1–20 Hours After Fever Onset (%)	Number of Animals that Survived/Total Number Treated 20.1–30 Hours After Fever Onset (%)	DT50 estimate
Ciprofloxacin	9/10 (90%)	8/10 (80%)	2/7 (29%)	19.7a hours
Levofloxacin	6/6 (100%)	9/9 (100%)	8/14 (57%)	26.5b hours

Abbreviation: CI, confidence interval; DT50, the time between fever onset and the initiation of antimicrobial treatment that results in 50% survival of the animals.

^aDT50 was 19.7 hours (95% CI, 12.0–43.0) by logistic regression that predicts survival using time from fever to treatment, with the confidence interval calculated by using asymptotic normality of the parameters with Monte Carlo simulation.

^bDT50 was 26.5 hours (95% CI, 22.7–32.1), using the same methods as with the ciprofloxacin study.

Figure 1.

The probability of death is plotted versus the time after fever in which ciprofloxacin treatment is initiated. The dots are observed deaths (probability of death = 1) or survivors (probability of death = 0). The solid gray line is the predicted probability of dying from the logistic regression, with a main effect for time from fever to treatment, and the dotted gray lines give the 95% pointwise confidence intervals on the predicted probabilities.

Levofloxacin Study

In this study, 39 AGMs were challenged with the CO92 strain of Y. pestis via an inhalational route (Table 1). Challenge doses ranged from 41 LD₅₀s to 160 LD₅₀s. The average challenge dose was 92 LD₅₀s. The AGMs were randomized to be treated with levofloxacin (or placebo) at various times after the onset of fever. The time of fever onset ranged from 52.5 to 78.4 hours (data not shown). The time to fever onset was negatively correlated with the challenge dose (r = −0.35; 95% CI, −.60 and −.04). Placebo was administered to 8 AGMs (5% dextrose). All 8 placebo animals died within 40 hours of the time of fever. The time to death was nonsignificantly, negatively correlated with the challenge dose (r = −0.49; 95% CI, −.89 to .32). There was 1 placebo animal that died prior to the initiation of treatment and 1 that was inadvertently given placebo prior to the onset of fever. There was 1 AGM that died prior to the first scheduled levofloxacin dose and 2 that were inadvertently treated prior to the onset of fever (Table 1). The remaining 29 AGMs were treated at various intervals after the onset of fever (Table 1). The 29 animals treated with levofloxacin after the onset of fever demonstrated that a delay in treatment results in a reduction in efficacy (Table 2). The statistical analysis of this data revealed that delaying levofloxacin treatment by 26.5 hours results in a treatment efficacy of 50%, as compared to the higher level of efficacy observed when treatment is initiated at or near the time of fever (Table 2). Figure 2 depicts the clinical trajectory of the 29 individual AGMs treated with levofloxacin after the onset of fever and a curve fit providing an estimate of rate of efficacy decay.

Figure 2.

The probability of death is plotted versus the time after fever in which levofloxacin treatment is initiated. The dots are observed deaths (probability of death = 1) or survivors (probability of death = 0). The solid gray line is the predicted probability of dying from the logistic regression, with a main effect for time from fever to treatment, and the dotted gray lines give the 95% pointwise confidence intervals on the predicted probabilities.

DISCUSSION

This work demonstrates that the efficacy of both ciprofloxacin and levofloxacin are limited to a narrow time window following the onset of fever. The statistical analysis of the ciprofloxacin data estimates that treatment initiated 19.7 hours after fever onset would lead to the survival of 50% of AGMs, while levofloxacin treatment initiated 26.5 hours after fever would lead to the survival of 50% of AGMs. The shortest delay in ciprofloxacin treatment that resulted in death was 7.5 hours postfever, while for levofloxacin the shortest delay was 24 hours; conversely, the longest delays that resulted in survival were 28.5 hours for ciprofloxacin and 27.5 hours for levofloxacin. It is not clear what explains the differences observed for the 2 studies and/or antimicrobials, or whether these differences are significant and/or reproducible.

Lower challenge doses were achieved in the levofloxacin study, which led to differences in the time to fever between studies (P < .001 by the Wilcoxon-Mann-Whitney test). The differences in the challenge doses may also have led to differences between the 2 placebo groups in the different studies; the placebo animals in the levofloxacin study lived an average of 40 hours postfever before succumbing to plague, while placebo animals in the ciprofloxacin study lived an average of 20 hours postfever before succumbing to plague (data not shown).

Perhaps a more likely explanation is that a faster in vitro rate of killing by levofloxacin, compared with ciprofloxacin, has been in observed in some studies [5–7], although not all studies [8].

Finally, it is possible that higher exposures of levofloxacin (area under the curves) were achieved due the differences in the dosing schemes. These differences reflected efforts to model the human doses of these fluoroquinolones. Levofloxacin was administered in a split daily dose of 8 mg/kg, followed 12 hours later by 2 mg/kg, to approximate the pharmacokinetic profile seen in once-daily dosing of humans. With ciprofloxacin, equivalent doses were given every 12 hours, as in human dosing. In vitro data supports the notion that dosing differences could explain why levofloxacin was more effective than ciprofloxacin when treatment is initiated at a later time [9].

CONCLUSIONS

This work has established a relationship between the time of treatment initiation and the efficacy of 2 FDA-approved fluoroquinolone antimicrobials in the treatment of pneumonic plague in an AGM model. Given the established similarities between the human disease course and the AGM disease course, this is likely to provide useful information to clinicians and argues for prompt antimicrobial therapy in humans with plague. In addition, this work provides information for designing studies to evaluate therapies that have the potential to increase the efficacy of those antimicrobials given after the time of fever initiation in AGMs and humans.

Notes

Acknowledgments. All authors thank Battelle Memorial Institute, Hank Lockman, Roy Barnewall, and Susan Houser for performing the ciprofloxacin work and the Lovelace Biomedical and Environmental Research Institute, William Mega, Robert Sherwood, and Philip Kuehl for performing the levofloxacin work.

Financial support. This work was supported by the National Institute of Allergy and Infectious Diseases, part of the National Institutes of Health (ciprofloxacin work, contract number HHSN272201000003I and task order number HHSN27200009; levofloxacin work, contract number HHSN272201000017I and task order number HHSN27200005).

Supplement sponsorship. This article appears as part of the supplement “Plague and Bioterrorism Preparedness,” sponsored by the Centers for Disease Control and Prevention.

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.